“The functional data are not useful until you can establish a clinical link.” — Marcy Richardson, PhD, Associate Director at Ambry Genetics
As Associate Director at Ambry Genetics, Marcy Richardson, PhD, has spent the last few years focusing on materializing precision medicine by making sense of the wealth of clinical genomics data the company has gathered over the past couple of decades.
Still, sorting through the millions of variants each individual possesses (compared to human reference genomes) and identifying the true pathogenic variants that cause a functional change has always been a challenge with human genetics data. Apart from the few glaringly obvious pathogenic variants, most end up receiving the blanket “variant of unknown significance” (VUS) label, effectively putting clinical genomics data sets to gather dust on a hard drive or cloud and leaving patients without any more answers than when they came into the clinic to be sequenced as precious time ticks away.
Take the DNA binding domain (DBD) of the infamous breast cancer gene BRCA2—a stretch where 557 missense variants of the possible 6,960 variants (8%) have been functionally classified.
Directly across from Richardson’s vantage point is that of the researcher performing high-throughput screens, pinning down variants that cause a deleterious effect, such as dysregulated cell growth or cell death. However, without being merged with clinical data, all of the variants from these screens are in purgatory between lab experiments and real patient data.
“You can do functional studies all day long, but until you can say that this is a clinically meaningful result, they’re not useful to the clinical teams,” Richardson told Inside Precision Medicine. “If you have a pathogenic mutation and there are many functional studies that have a whole group of non-functional variants, but none of those variants are actually pathogenic, they do not cause disease, and we do not necessarily understand why, but without that clinical link, you cannot use that information to interpret variants. You have to say this category of variants is seen in people who have a higher incidence of cancer XYZ, and then once you establish that link, you can say, ‘Let’s go back and use that functional data to interpret the variance.’”
Sequencing and gene synthesis advances have made feasible multiplexed assays of variant effect (MAVEs), which quantify the functional impact of many thousands of genomic variants in a single experiment. These assays and the functional evidence they generate potentially empower more accurate clinical variant classification. While there have been MAVE studies of BRCA2, they have been limited to proof-of-principle efforts focusing on relatively small regions of BRCA2 and lacking validation.
In collaboration with the lab of Fergus J. Couch, PhD, Professor and Chair of the Division of Experimental Pathology and Laboratory Medicine at the Mayo Clinic, Richardson and Ambry were able to intersect their clinical genomics data with results from a CRISPR-Cas9-based saturation genome editing (SGE) screen of evaluating every single possible variant in the BRCA2 DBD.
“Once we opened this door to MAVE experiments, it was an inevitable next step for BRCA2, a gene that impacts many people,” said Richardson. “BRCA2 pathogenic variants carry a high risk for breast and ovarian cancer, and it’s a sizable gene, meaning there are a lot of VUSs—and nobody likes a VUS.”
According to research published today in Nature, by integrating the BRCA2 SGE results with three clinical classification frameworks (ClinGen, ACMG, and AMP), the Richardon and Couch collaboration, led by co-first authors Huaizhi Huang and Chunling Hu, was able to assign 91% of variants as pathogenic/likely pathogenic or benign/likely benign.
“There are all kinds of shades of gray where the variants that behaved the worst had the highest incidence of breast and ovarian cancer, and then the next worst had slightly less breast and ovarian cancer, so it was really beautifully demonstrated with the clinical data that we were able to share.”
This comprehensive classification significantly enhances our understanding of BRCA2 mutations, their potential impact on cancer risk, and the ability to manage individuals with BRCA2 variants by providing more explicit clinical guidance. These data will prove helpful in the future, through integration with other datasets, for the characterization and classification of all variants in this genetic location in individuals from all racial and ethnic backgrounds and all BRCA2-associated forms of cancer.
According to Richardson, this information is immediately clinically impactful, especially for those waiting for therapies, and can be added to a patient’s variant interpretation in their genetic test report. “Now, they don’t have to wait months before they find out that they have a pathogenic variant, and by then they’ve already missed the vote on heart inhibitor therapy,” said Richardson.
Richardson’s statement is supported by the co-publication of this Nature paper alongside an article from Shyam K. Sharan, PhD, at the National Cancer Institute (NCI), which yielded comparable results, though the precise overlap in the results remains unclear. The main difference between the articles is that Sharan’s group used a humanized mouse embryonic stem cell line instead of the haploid human cell line used by the Mayo Clinic and Ambry Genetics collaboration.
“It’s really nice when the timing of things comes together in this way because everyone gets the credit, and the community gets double the data,” said Richardson. “The fact that these were co-published in this journal will set the bar for how we interpret these data in the future because large-scale studies like this are inherently more messy.” He continued saying that having both data sets available will facilitate translating the data to patient applications and reduce the bottlenecks inherently associated with interpretation and analysis.
The approach taken here by Ambry Genetics and the Mayo Clinic is also a small step toward progress for healthcare equity. Richardson explained, “The old model was once we’ve seen a variant in a patient, then let’s go and study that variant and resolve it. But who’s getting tested to start with? It’s not the underserved populations. Previously, we were only studying variants that showed up in white people. This study provides us with readily available data, and the first time we see a variant in anyone, it is to be able to have information for basically every single variant that we’ve never seen yet.”
